Temporal Weighted Association Rule Mining for Classification

Purushottam Sharma, Kanak Saxena
2012 Journal of clean energy technologies  
There are so many important techniques towards finding the association rules. But, when we consider the sale of seasonal items (Winter, Summer, Spring etc.) or weighted items then no effective algorithm or model existed till now that can able to mine the interesting pattern on time-variant seasonal database. In view of this, we propose an optimized Temporal Weighted Association Rule miner (abbreviatedly as TWARM) algorithm to perform the mining for these problems and we conduct the
more » ... ct the corresponding performance studies by implementing this algorithm and then we used these weighted association rules for designing a good classifier to classify the items. Furthermore, without fully considering the time-changing characteristics or behavior of items and transactions, it is noted that some discovered rules may be expired from user's interest i.e. rules generated in one season can not give the useful and required information in other season. Under TWARM we first partition the database on the bases of seasons (winter, summer, spring) or Yearly, Half Yearly or Quarterly etc. according to user's requirements and then we apply temporal weighted mining on each partition. In TWARM (Temporal Weighted Association Rule Miner) the cumulative occurrence count of mining previous partitions is selectively carried over toward the generation of candidate itemsets for the subsequent partitions. We have also applied scan reduction technique in TWARM due to that only two scan of database are required, means saving lots of time. Now we have all Temporal Weighted Association Rules for Classification (TWARC) with the help of TWARM. By using these Temporal Weighted Association Rules we design a classifier to classify the items towards the appropriate class symbol. Index Terms-Weighted transaction, time weighted mining, seasonal mining, temporal weighted association rule miner, Classification based on weighted association.
doi:10.7763/ijcte.2012.v4.585 fatcat:64w7txnbnrdx3gthr5kkmzvcoq